Guiding analysts and rational autocorrelation in forecast error

Empirical investigations of analyts forecast surveys concerning financial or macroeconomic variables find significant time varying biases such as autocorrelation in the time series of forecast errors. This fact is usually attributed to behavioural biases of analysts. We develop a random dynamical system describing the evolution of forecasts, actual outcomes and fundamental values, when insiders manipulate analysts forecasts to minimize short-term variation in the actual outcomes. Studying optimal control, we can show that autocorrelated forecast errors are induced and, hence, rational. The Reuters survey of institutional EUR/USD forecasts reveals indication, that the ECB has tried to control the volatility of the external value of Eurolands assets. However, volatility in the exchange rate is rather increased and serially correlated forecast errors are created.